Vol. 406: 1–17, 2010
doi: 10.3354/meps08588
MARINE ECOLOGY PROGRESS SERIES
Mar Ecol Prog Ser
FEATURE ARTICLE
Published May 10
OPENPEN
ACC E S
SCCESS
Modelling the effects of chromatic adaptation on
phytoplankton community structure in the
oligotrophic ocean
A. E. Hickman1,2,* , S. Dutkiewicz2, R. G. Williams1, M. J. Follows2
2Department
1Department of Ocean and Earth Sciences, University of Liverpool, Liverpool L69 3GP, UK
of Earth, Atmospheric and Planetary Sciences, Massachusetts Institute of Technology, Cambridge, Massachusetts 01239,
USA
© Inter-Research 2010 · www.int-res.com*Email: ahickman@liv.ac.uk
ABSTRACT: We explored the role of chromatic adaptation in
shaping vertical phytoplankton community structures using a
trait-based ecosystem model. The model included 1000
‘phytoplankton types’ and was applied to the oligotrophic South
Atlantic Gyre in a 1dimensional framework, where
‘phytoplankton types’ refers to the model phytoplankton that were
stochastically assigned unique physiological characteristics. The
model incorporates multi-spectral optics and light absorption
properties for the different phytoplankton. The model successfully
reproduced observed vertical gradients in the nitrate, bulk
phytoplankton properties and community structure. Model
phytoplankton types with Synechococcus-like spectral light
absorption properties were outcompeted at depth, where
eukaryote-like spectral properties were advantageous. In contrast,
photoinhibition was important for vertical separation of high-light
and low-light Prochlorococcus model analogues. In addition,
temperature dependence was important for selection of
phytoplankton types on the temperature gradient. The fittest, or
successful, phytoplankton types were characterised by
combinations of simultaneously optimal traits that suited them to
a particular depth in the water column, reflecting the view that
phytoplankton have co-evolved multiple traits that are
advantageous in a particular environmental condition or niche.
KEY WORDS: Phytoplankton · Chromatic adaptation ·
Photosynthesis · Niche · Modelling · Species · Selection ·
Pigments
Resale or republication not permitted without written consent of the publisher
The underwater light spectrum provides a complex environment for
phytoplankton photosynthesis and growth.
Image: Alex Mustard (amustard.com)
INTRODUCTION
Phytoplankton community composition varies widely over the
global ocean resulting from differences in fitness and species
selection in the physical, chemical and predatory environment.
Relative fitness of phytoplankton is determined by a number of
factors, including the growth dependence on nutrient
availability, light and temperature, as well as susceptibility to
predation, viral lysis and other causes of mortality.
2
Mar Ecol Prog Ser 406: 1–17, 2010
In the open ocean, oligotrophic environment of the extensive
subtropical gyres, there is distinct vertical structuring of the
habitats of the pico-phytoplankton (Fig. 1): Synechococcus
usually occupy surface waters,
whilsteukaryotesandProchlorococcus occurthroughout the
water column, normally with subsurface biomass maxima
(Partensky et al. 1996, Zubkov et al. 2000). Vertical gradients
among different Prochlorococcus ecotypes have also been
observed in oligotrophic regions (Moore et al. 1998, Johnson et
al. 2006). In these stratified water columns, the vertical
gradients in phytoplankton community structure are coincident
with gradients in temperature, nutrient and light availability.
Our aim was to explore how these environmental factors act in
forming the observed phytoplankton distributions. We
examined the hypothesis that, through their pigment
compositions, the spectral light absorption characteristics of
different phytoplankton are important in determining species
selection in the spectral light gradient (Bidigare et al. 1990a,
Moore et al. 1998, Sathyendranath & Platt 2007, Hickman et
al. 2009).
05
Syn 0 5
0401
0401
0
-100
-200
10 -300
0
10
OBSERVATIONAL CONTEXT
40°S 30°W
30°E
20°S
0
We focussed our study on the South Atlantic Gyre, which is
05
broadly representative of much of the oligotrophic ocean (Fig. 1).
The physical environment is relatively stable and has relatively
weak temporal variability. We used a trait-based modelling
approach (Follows et al. 2007) that was modified to include
multispectral optics and light absorption characteristics for
different phytoplankton. In the model, traits were assigned
stochastically for a large number of ‘phytoplankton types’, where
each phytoplankton type has a unique combination of parameter
values that govern its growth dependencies with respect to light,
nutrients and temperature. In this framework, phytoplankton
community structure is an emergent property resulting from the
interaction, competition and selection of phytoplankton types in
the virtual ocean environment. We explored the model outcomes
and a series of model ‘thought experiments’ to investigate, in
particular, the role of the spectral composition of in situ
irradiance and phytoplankton light absorption, along with nutrient
and temperature dependence on growth, in shaping the habitats of
the fit and abundant phytoplankton types.
10 0
Nitrate
(µM)
20°N
Our study complements the work of Stomp et al. (2004, 2007),
who illustrated the importance of spectral light absorption
properties on the coexistence of 2 species of Synechococcus
in laboratory cultures and an idealised model. Previous
modelling studies have also explored the importance of
differences in photophysiology for niche separation of
Synechococcus and Prochlorococcus (Rabouille et al.
2007), and have incorporated spectral light absorption
properties in global phytoplankton functional type models (e.g.
Gregg & Casey 2007). Here, we used an ecosystem model to
explicitly explore the role of different traits, specifically the
nutrient, temperature and spectral light dependencies, in the
competition among and selection for different phytoplankton
taxa, and thus in forming observed distributions.
0°
) of Synechococcus (Syn),
Prochlorococcus (Pro) and
picoeukaryotes (PEuk) along with
corresponding temperature and nitrate
profiles at (d) 3 locations in the South
Atlantic Gyre. Data were obtained on (a)
14, (b) 15 and (c) 17 October 2004 (nitrate
in Panel c from 15 October). Analytical
flow cytometry measurements provided by
J. Heywood and M.
Zubkov
PEuk 0
(c
)
To the latter represent a mixed group (d)
provid
of small eukaryotes (<~2 µm).
e a Abundances of these cells were
context
measured by analytical flow
for the
cytometry following methods of a b
study,Heywood et al. (2006) and using c
we the constant biomass conversion
first factors of
consid
ered
the
phytop
lankto
n
distrib
utions
observ
ed in
the
South
Atlanti
c Gyre
(Fig.
1).
Data
were
collect
ed on
the
Atlanti
c
Meridi
onal
Transe
ct
(AMT15
from
the UK
to
Cape
Town,
South
Africa,
17
Septe
mber
to 29
Novem
ber
2004),
includi
ng
inform
ation
on the
vertica
l
distrib
utions
of
Prochl
oroco
ccus,
Synec
hococ
cus
and
picoeu
karyot
es,
where
15 25 Temp (°C)15 25 15 25
Hickman et al.: Modelling phytoplankton communities
Zubkov et al. (1998). These small cells accounted for ~80% of the
phytoplankton chlorophyll a (chl a), as measured by size
fractionation. Synechococcus were restricted to the surface
mixed layer (Fig. 1), whilst Prochlorococcus and picoeukaryotes
together dominated the biomass throughout the water column.
Both Prochlorococcus and picoeukaryotes exhibited subsurface
biomass peaks. The vertical gradients in the phytoplankton
community structure are broadly persistent over the South
Atlantic Gyre (Fig. 1), and are in accord with previous
observations in the region (Zubkov et al. 2000), and those in other
oligotrophic regions (Partensky et al. 1996). Vertical gradients in
different Prochlorococcus ecotypes have also been observed in
the South Atlantic Gyre (Johnson et al. 2006).
A variety of differences between Synechococcus, eukaryotes
and Prochlorococcus ecotypes may contribute to their
interactions and species selection, including their distinct pigment
compositions. Specifically, and in addition to other pigments,
Synechococcus contain phycobiliproteins that absorb light
between 495 and 620 nm, whilst eukaryotes contain various
photosynthetic carotenoids that absorb light of 490 to 510 nm
(Bidigare et al. 1990b, Jeffrey et al. 1997; Fig. 2). Only a minority
of eukaryotes also contain phycobiliproteins (Jeffrey et al. 1997).
Similarly, Prochlorococcus has high-light and low-light
ecotypes that differ in their pigment ratios (particularly divinyl chl
b:divinyl chl a), leading to preferential absorption of light at
different wavelengths (Moore et al. 1995; Fig. 2).
3
Vertical gradients in the phytoplankton community structure are
reflected in pigment concentrations and lead to shifts in relative
light absorption with depth (Babin et al. 1996, Barlow et al.
2002). For the South Atlantic Gyre, the ratio of phytoplankton
light absorption at 490:550 nm increased from 0.62 at the surface
to 0.73 at depth, reflecting a shift in phytoplankton light
harvesting from the green to blue part of the light spectrum. Light
absorption measurements during AMT-15 were made following
Hickman et al. (2009) based on the methods of Bricaud &
Stramski (1990).
MODEL
The multi-species model was based on that of Follows et al.
(2007) and was applied in a 1-dimensional framework for a
location in the South Atlantic Gyre (17.5° S, 24.5° W). The model
was significantly modified so that light dependence of growth for
the different phytoplankton types was determined from assigned
light absorption spectra from culture data, and a parameterisation
of photophysiology was introduced, based on Geider et al. (1997).
In the following sections, we briefly describe relevant aspects of
the model and the modifications.
Physics and biogeochemistry. The foundation of the
stochastic phytoplankton-modelling framework, and the
parameterisations therein, were described by Follows et al.
(2007) with modifications as in Dutkiewicz et al.
Fig. 2. Chlorophyll (chl) a-normalised light absorption spectra for 4 different phytoplankton
types. HL: high-light; LL: low-light. Measurements were made on cultured phytoplankton
under similar growth irradiances. Solid lines show light absorbed only by photosynthetic
pigments (achlps chl(λ)); dashed lines show the light absorbed by all pigments (a(λ)). Difference
between the solid and dashed lines subsequently illustrates the amount and spectral distribution
of light absorbed by the non-photosynthetic carotenoids. There was a negligible amount of
such pigments in the eukaryotes. Culture data provided by L. Moore and D. Suggett
– 5 2m– 1 s(2009).
Here the model was
configured to resolve only the vertical
dimension, with a resolution of 10 m from
the surface to a depth of 150 m, 20 m for the
next 100 m, and gradually increasing
thickness with depth thereafter. Mixed layer
deepening and shoaling were achieved by a
simple convective adjustment
parameterisation driven by a relaxation of
sea surface temperature towards the
climatological seasonal cycle on a timescale
of 3 d (World Ocean Atlas; Stephens et al.
2002). A background diapycnal diffusivity
of 10was also imposed,in accord with
observations from the subtropical
thermocline (Ledwell et al. 1993). As in
Dutkiewicz et al. (2009), the model includes
the cycling of nitrogen and phosphorus (as
well as iron and silicate) and a relatively
simple parameterisation of the
remineralisation of organic matter by
heterotrophic microbes and assumes fixed
Redfieldian elemental ratios in
phytoplankton
4
Mar Ecol Prog Ser 406: 1–17, 2010
(120:16:1 C:N:P). Given the focus of the current study, no large
phytoplankton types were initialised, no types required silicate,
and iron concentrations were maintained so that iron was never
limiting.
A climatological annual cycle of 24 h mean irradiance for the
South Atlantic Gyre location, obtained from Sea-viewing Wide
Field-of-view Sensor (SeaWiFS) photosynthetically available
radiation (PAR) data, was imposed at the surface. Initial nutrient
fields were obtained from the World Ocean Atlas 2001 (Conkright
et al. 2002), and the delivery of nitrate and phosphate due to
lateral advection and isopycnal transfer, important in oligotrophic
gyres (Williams & Follows 1998, Dutkiewicz et al. 2005), was
parameterised by additional sources throughout the upper 200 m.
The imposed convergence of lateral fluxes (0.00125 µM P yr– 1
and 0.02 µM N yr– 1 ) were consistent with those inferred from
climatological data (Williams & Follows 1998) and as diagnosed
in a 3-dimensional model (Dutkiewicz et al. 2009).
phytoplankton type at each of the 13 wavelengths, and chlj(zk– 3 )
is the chl a concentration (mg chl a m) for each phytoplankton
type for the depth bin zk. The irradiance used for calculating
phytoplankton growth was the mean irradiance within the depth
bin:
( ,
)
e x p
.
l n
( ,
I
z
I
z
λ
λ
=
[
]
Total PAR was obtained from the sum of
irradiancek
)
l n
( ,
( I
– 2 – 1 s(Suggett
et a
m(Suggett et al. 20
z
λ
+
(3)
k
k
between 400 and 700 nm, assuming a linear interpolation between
the 13 wavelengths.
Spectral irradiance. The ecosystem model here
differed from the previously published configurations
(Follows et al. 2007, Dutkiewicz et al. 2009) in terms of
the treatment of light and photophysiology. Here we
resolved the wavelength composition by considering
irradiance at 13 wavelengths between 400 and 700 nm,
more than the minimum of 6 recommended by Kettle &
Merchant (2008). The (normalised) wavelength
spectrum of incident irradiance was obtained from
optics profile data collected in the South Atlantic Gyre
region during AMT-15 (via extrapolation of the light
intensity at available wavelengths to the surface), and
was scaled by the incident PAR to spectrally resolve
incident irradiance. The seasonal change in the incident
light spectrum was not considered, but is minimal
compared to the variation with depth.
– 2 – 1 sPhytoplankto
n light absorption. Phytoplankton light absorption
spectra were obtained from cultures of 4 species
grown at similar growth irradiances, which were
considered to be representatives of high-light
Prochlorococcus (HLPro), low-light
Prochlorococcus (LLPro), Synechococcus (Syn)
and eukaryotes (Euk). The cultures were:
Prochlorococcus MED4 and SS120 grown at 70
µmol photons m– 2 – 1 s(Moore & Chisholm 1999),
Synechococcus WH7803 grown at 50 µmol
photons m
–2 –1
Penetration of irradiance, I (µmol photons m
I
z
, ) ( , )exp[ ( ,
)(λ
λ
s
zzk / )
]
k
I
z
k
z
z
(
λ
1 2
1 2
1 2 d
k
k
+
+
/
(λ))
non
wav
200
obta
foll
tech
200
k
k
/
/
- 1 2
)
for each of the 13 wavelengths, λ (nm), was obtained
at each depth, z(m),
(1)=
-
The light absorption by all pigments provided by the culture
measurements (aj chl(λ)) was used in the calculation of light
attenuation, whereas estimation of phytoplankton growth was
where zk–1/2
and zk+1/2
r
e
f
e
r
t
o
t
h
e
d
e
p
t
h
s
o
f
t
h
e
t
o
p
a
n
d
b
o
t
t
o
m
o
f
t
h
e
d
e
p
t
h
b
i
n
r
based on light absorption only by photosynthetically active pigments (aps,j
was obtained by ad
chl(λ)), thus ignoring absorption by non-photosynthetic carotenoids (Fig. 2).
spectra (aj chl
The light absorption spectrum for only the photosynthetic pigments (aps,j chl(λ))
e
s
p
e
c
t
i
v
e
l
y
,
a
n
d
s
u
b
s
c
r
i
p
t
k
i
s
c
t
r
a
l
l
i
g
h
t
a
t
t
e
n
u
a
t
i
o
n
,
k
)
,
w
a
s
a
c
o
u
n
t
e
r
f
o
r
t
h
e
d
e
p
t
h
b
i
n
.
T
h
e
s
p
e
c
a
l
c
u
l
a
t
e
d
f
r
o
m
t
h
e
a
t
t
e
n
u
a
t
i
o
n
b
y
water, k(λ) (Pope & Fry 1997), colored
organic matter (CDOM), k(λ) (from Kiti
using coefficients from the South Atlant
region) and phytoplankton,
T
h
e
t
e
d
r
e
l
a
t
i
v
e
b
y
p
i
g
m
e
n
t
c
o
n
c
e
n
t
r
a
t
i
o
n
s
f
o
r
t
h
e
r
e
c
o
n
s
t
r
u
c
t
i
o
n
s
w
e
r
e
c
a
l
c
u
l
a
s
c
a
l
i
n
g
t
h
e
w
e
i
g
h
t
s
p
e
c
i
f
i
c
a
b
s
o
r
p
t
i
o
n
s
p
e
c
t
r
a
f
o
r
t
h
e
m
a
i
n
p
i
g
m
e
n hl a, b and c and phycoerythrobilin-rich
t relevant for the representative Synechoc
Bidigare et al. 1990b) expected for each
g type (according to Jeffrey et al. 1997) in
r the lowest sum of residuals between the
o pigment-reconstructed and measured ligh
u spectra. For each of the 4 phytoplankton
p average residuals across all wavelengths
s m
(
p
h
o
t
o
s
y
n
t
h
e
t
i
c
c
a
r
o
t
e
n
o
i
d
s
,
n
o
n
p
h
o
t
o
s
y
n
t
h
e
t
i
c
c
a
r
o
t
e
n
o
i
d
s
,
c
where the attenuation by phytoplankton was summed over all
phytoplankton types, j. aj chl2mg– 1 chl a) for each(λ) is the chl
anormalised light absorption (m
Phytoplankton growth. The growth rate, Pj C(s– 1 ), of each
phytoplankton type, j, was determined by the steady state
solution of the light-growth dependency of
Hickman et al.: Modelling phytoplankton communities
5
Ekjis the light saturation parameter and represents the intercept of
the maximum (PC m,j) and the linear initial slope [θjfma–––––––
(λ)] of the carbon-specific photosynthe-chl ps,j
sis versus irradiance curve,
=m Cj ja,θ f λ ( )
Geider et al. (1997), which provided a variable chl a
concentration, and was thus a modification of Follows et al.
(2007). The light-growth dependency was based on an
exponential form of the carbon-specific photosynthesis versus
irradiance curve, modified to resolve the visible spectrum,
(9)
Ek P
m
p
s
c
h
l ,
j
j
,
θ θj j= +Λm m1
2 Iθ
Fo
r
ph
yto
pla
nkt
on
typ
es
ass
ign
ed
to
be
sus
ce
pti
ble
to
ph
oto
inh
ibi
tio
n,
Pj
Cw
as
red
uc
ed
de
pe
ndi
ng
on
the
rati
o
of
irr
adi
an
(6)
1(10)j
T
)
(
)
T
B=
T
T
–
(
ce,
I
(th
e
tot
al
bet
we
en
40
0
an
d
70
0
nm
)
to
Ekj
–2(
µ
mo
l
ph
oto
ns
m–
1s
=
P
j
P
P
(4)
where achl ps,j––––––– (λ) is the mean light absorption by photo-
j
I jm Cj Ce x p 1 Λ θ ⎡
⎣⎛ ⎝⎜ ⎞⎢ ⎤m C⎠ ⎟⎦ ⎥ ,
j
where Pj Cis the carbon specific growth rate, and PC m,j(s– 1 ) is the
maximum (light saturated) photosynthetic rate, given by
(5) where µjis the maximum possible growth rate (s– 1 ), and γj
Nand γj T, reflect the degree of limitation by nutrients and
temperature and have values between 0 and 1 (described below).
The linear slope of the photosynthesis versus irradiance curve
depends on the chl a to carbon ratio,
θjPj
j
j
j
m
C
N
T
,= µ γ
γ (g:g), which
varies with photoacclimation (Geider et al. 1997):
Th
e
te
mp
era
tur
e-g
ro
wt
h
fun
cti
on
wa
s
bas
ed
on
an
Ep
ple
y
cur
ve
(E
ppl
ey
19
72)
,
synthetically active pigments across wavelengths of 400 to 700
nm. Photoinhibition only acts when irradiance is greater than Ekj.
The degree of photoinhibition was set by a factor κ, here equal to
1.2; for simplicity, there is no spectral dependence included for
photoinhibition.
The light-growth response for different phytoplankton types
was thus based on the assigned light absorption properties (achl
ps,j), along with an associated susceptibility (or not) to
photoinhibition.
wit
h
eac
h
ph
yto
pla
nkt
on
typ
e
ass
ign
ed
an
en
vel
op
e
wit
hin
wh
ich
it
gro
ws
eff
ici
ent
ly
(F
oll
ow
s
et
al.
20
07)
:
m C,j
γt
- j C
A exp
,
t
0
P
12
ΛI jin
Eqs. (4) and (6) is the instantaneous photosynthetic rate
given the amount of light absorbed by the phytoplankton, whichλmis
determined from the spectral irradiance and spectral light
absorption coefficient for each phytoplankton type,
=
The coefficients defining the temperature envelope were th
same for each phytoplankton type (A = 1.04, B = 0.001, C
t1= 3 and t2= 0.3, all dimensionless, as in Dutkiewicz et al.
2009), except T
p
∑
s
f
λ
λ
400a
)
,j
Λj =
where fm– 1 is the maximum quantum yield of carbon fixation
(mol carbon molphotons), and θis the maximum possible value I
of the chl a to carbon ratio. Both parameters were assumed to
be the same for all phytoplankton types (fmm– 1 = 0.04 mol
carbon molphotons and θm= 1/60 g:g, typical values observed
for phytoplankton; Kyewalyanga et al. 1998, Geider et al.
1997). Phytoplankton types were assumed to be fully
acclimated to in situ irradiance, as in the steady state solution
of Geider et al. (1997), which is reasonable given the daily
mean irradiance used and the relatively stable nature of the
oligotrophic environment. In addition, for the
Prochlorococcus-like phytoplankton types, divinyl-chl a was
assumed to be equivalent to chl a for all calculations.
Th
e
mo
del
inc
lud
ed
2
gra
λ=700chl
I
(
)
(
(7)
was determined by the most limiting nutrient (Follow
2007), with the individual nutrient limitations based
Michaelis-Menten function, Ni/(Ni+Kij), and modifie
nitrogen to include inhibition of nitrate uptake by am
(Dutkiewicz et al. 2009); subscript i indicates the nu
(phosphate, nitrate), Nis the nutrient concentration (µ
Kijiis the nutrient halfsaturation constant (µM). Nutri
dependencies differed between phytoplankton types
to the required sources of nitrogen and the halfsatura
constant, Kij, hereafter referred to as Ksat, which were
for N and P according to the fixed Redfield stoichiom
zer
s
(as
in
Du
tki
ew
icz
et
al.
20
09)
,
but
sin
ce
the
re
we
re
no
im
po
sed
siz
e
dif
fer
en
ces
bet
we
en
ph
yto
pla
nkt
on
typ
es
in
the
cur
ren
t
stu
dy,
gra
zin
g
did
not
inf
lue
nc
e
the
ph
yto
pla
nkt
on
co
m
mu
nit
y
str
uct
ure
.
P
P
E
k
j
j
j
n
h
i
i
b
C
C
λλ700400
∑=Iκ
,==λ
(
)
Trait choices. Trait choices for the model phytoplankton types ) according to (8)
were assigned based on a series of coin flips and random
selection of parameter values (Fig. 3). Phytoplankton types
were assigned to 1 of 3 nutrient resource groups: a requirement
for NH4only, NH4and
Phytoplankton type
25 50
25
NH4, N O 4
N
H
NH4, N O 2, N
O3
2
, NO
2
syn 3
sat
in combination, or NH
4
opt
syn,
sat
6
Mar Ecol Prog Ser 406: 1–17, 2010
eeuk, e
a
n
d
s
a
t
and eeuk
N
O
H
L
P
r
o
use
nitrate,
while e
5050505050
50
LLPro LLPro SynEukHLProHLPro
0.015 0.015
Ksat
0.035
0
.
Fig. 3. Framework of choices and parameter
0 ranges for growth characteristics during initialisation of the phytoplankton ty
was first assigned nitrogen resources and then allocated 1 of the 4 spectral light absorption properties (high-light Prochlo
1
Prochlorococcus [LLPro], Synechococcus [Syn] or eukaryote [Euk], as in Fig. 2), depending on the requirement for n
type was stochastically assigned a value for temperature optima Topt(°C) and half-saturation constant for nutrients, Ksat(µM
1 for Kdepend on the nitrogen resource). Values for Ksatare translated between phosphorus and nitrogen using
(again ranges
Numbers next
0 to tree branches indicate the probability (%) of each choice. Photoinhibition was imposed on all phytoplank
assigned light absorption properties of LLPro
T
o
p
t
K
3
0
; Eq. 10) and half-saturation coefficient (Kin combination, broadly
reflecting possible combinations for phytoplankton (Moore et al.
NO2 2002). Within each nutrient resource group, phytoplankton types
were assigned chl a-specific light absorption spectra according to
Figs. 2 & 3. Superimposed onto this framework was a random
selection for values for temperature optimum (T) from plausible
ranges (Fig. 3). The probability of trait choices was chosen so that
a roughly equal number of phytoplankton types were assigned
with each of the 4 different sets of light absorption properties
(Fig. 3).
The phytoplankton types that did not require nitrate were
assumed to be representative of Prochlorococcus and, thus, the
lower Krange was a reasonable assumption given their small size
(Chisholm 1992). Photoinhibition was imposed on phytoplankton
types that were assigned LLPro absorption characteristics as
suggested by culture studies (Moore et al. 1998, Six et al. 2007).
Model application and assessment. The model was initialised
with 1000 phytoplankton types, which ensured that the possible
trait parameter space was sufficiently well sampled to eliminate
any significant dependence of the solutions on the stochastic
initialisation. The model was integrated for 20 yr, after which
time the annual cycles of nutrient fields and phytoplankton
community structure were repeatable and consistent. All
phytoplankton types were retained
through
in each ensemble member was similar, and the ensemble mean
out the
was robust. The ensemble mean was obtained from the mean
modelbiomass of phytoplankton types when grouped according to light
runs, absorption properties and whether they use nitrate or not (Fig. 3).
even The
if notation eis used to refer to all phytoplankton types with Syn,
their Euk, HLPro and LLPro absorption properties, respectively, such
biomas
that eand eLLProdo not. We first assessed the skill of the model by
s wascomparvery ing the ensemble mean to observations collected during
low. AMT-15. Then we assessed the traits important to forming the
Twenty
community structure. This assessment was done by considering
differe
model outcomes as well as using the model to conduct a series
nt of ‘thought experiments’. Appendix 1 contains a suite of
indepe
thought experiments that sequentially illustrate the roles of the
ndentlight, nutrient and temperature dependencies in turn on the
integrat
community structure. All model outcomes shown are for
ions Month 10, in order to compare to AMT-15 data.
(‘ense
mble
membe
rs’)
were
conduc
ted,
each
with a
unique
stochas
tic
initialis
ation of
phytopl
ankton
types.
While
each
ensemb
le
membe
r
resulte
d in a
slightly
differe
nt
emerge
nt
commu
nity
structur
e (a
selectio
n is
shown
in Fig.
4),
importa
ntly,
the
broad
charact
er of
the
vertical
biogeo
graphy
Hickman et al.: Modelling phytoplankton communities
7
Fig. 4. Vertical profiles of carbon biomass for all 1000 phytoplankton types for 5 of the 20 ensemble members. Individual phytoplankton types are
identified by colour according to their light absorption properties (Syn: Synechococcus; Euk: eukaryote; HLPro: high-light Prochlorococcus; LLPro:
low-light Prochlorococcus). Multiple lines of the same colour therefore represent phytoplankton types with the same light absorption properties, but
different values for the half-saturation constant (K) and the optimum temperature for growth (Topt)sat
RESULTS
Nitrate, biomass and chlorophyll
We first compared the detailed hydrographic and biological
gradients measured at one of the stations in the South Atlantic
Gyre during AMT-15 (corresponding to Fig. 1c) to the mean
ensemble outcome (Fig. 5). The model represented a shallower
mixed layer than indicated in the (pre-dawn) observations (Fig.
5a,b), due to the monthly mean model output. Simulated nutrient,
biomass and chl a profiles were consistent with the observations
(Fig. 5a,b). The model successfully reproduced a deep chlorophyll
maximum (DCM) coincident with the depth of the nitracline. The
modelled absolute chl a concentrations were lower than observed,
but the total carbon biomass for the phytoplankton was about the
same as in the data at the surface as well as at the DCM, perhaps
reflecting the assumption of instantaneous acclimation of
chlorophyll to carbon ratios.
Phytoplankton community structure
The model qualitatively captured the observed gradients in
community composition. The coexistence of phytoplankton types
that use nitrate (assumed to represent eukaryotes and
Synechococcus) and those that do not (assumed to represent
Prochlorococcus) was reproduced in the model. Within the
subset of nitrateusing phytoplankton types, the biomass maximum
for those with Syn absorption spectra, esyn, was shallower
in the water column than those with Euk absorption spectra, eeuk,
consistent with the distributions of Synechococcus and
picoeukaryotes in the data (Fig. 5c,d). Similarly, for the
non-nitrate using types, the biomass maxima for those with
HLPro absorption spectra, e, was shallower than for those with
LLPro absorption spectra, eLLPHLP. While the vertical structuring
of niches seemed
reasonable, the relative biomass of the different phytoplankton
types differed, sometimes significantly, from that in the
observations. For instance, the biomass of esynwas
overestimated and ewas underestimated at the surface (such
that the biomass for esynwas greater than eeukeuk), although at the
DCM the biomass for each type was reasonable. Conversely,
the combined biomass of eLLPand eHLPin the model was similar
to the biomass of Prochlorococcus at the surface, but
overestimated at the DCM. However, the important features
reproduced by the model are: the surface bias of phytoplankton
types with Syn absorption spectra, and their absence at depth;
the ubiquity of phytoplankton types with Euk absorption
spectra, and their biomass maximum at depth; the vertical
separation of phytoplankton types with HLPro and LLPro
absorption spectra; and the coexistence of the nitrate and
non-nitrate using phytoplankton types. A traditional, after the
fact ‘tuning’ of model parameters could in theory improve the
quantitative comparisons, but this is not the philosophy of this
self-assembling community approach. We therefore proceed by
considering these qualitatively realistic and robust features of
the vertical distributions.
8
Mar Ecol Prog Ser 406: 1–17, 2010
Chl a (mg m–3)
Nitrate (µM)
0 0.2
0 0.1
0 0.2 Syn
0 2 Syn
00 3 PEuk
0 10
0 (a)
(c)
0 10
0 3 Euk
(d)
(b)
100
100
200
200
Temp (°C)
10 25 300
0 20 PEuk
0 2 Pro
30015 25
0 0.02 Euk
(e)
Fig. 5. (a,c,e) Data
from the South
Atlantic Gyre
compared to (b,d,f)
the mean of all model
ensembles. Data are
vertical profiles of (a)
temperature, chl a
and nitrate, (c)
carbon biomass (mg
C m– 3 ) of
Synechococcus
(Syn),
Prochlorococcus
(Pro), and
picoeukaryotes
(PEuk) estimated by
analytical flow
cytometry and (e) the
ratio of red (chl a)
fluorescence to
particle side scatter
(which correlates to
particle size) as
measured by flow
cytometry (arbitrary
units); this ratio is an
indicator of the chl
ato carbon ratio for
Synechococcus,
Prochlorococcus
and picoeukaryotes.
Model outcomes are
for the ensemble
mean (solid lines) ±
SD (dashed lines),
where phytoplankton
types were grouped
according to their
light absorption
properties. Outcomes
are (b) temperature,
chl a and nitrate, (d)
carbon biomass (mg
C m) for
phytoplankton types
with Syn, Euk,
HLPro and LLPro
–3
(f)
0
0 5 Pro
100
200
0 0.02 HLPro
300
0
L
L
L
P
r
o
0 50 Syn
Ratio chl a :C
Depth (m)
Depth (m)
Photophysiology of phytoplankton types
The ratio of phytoplankton chl a to
carbon increased with depth in the
model as a result of photoacclimation
(Geider et al. 1997), as observed in
the data (Fig. 5e,f). The available light
spectrum became depleted in
irradiance at 550 nm relative to 490
nm with depth, as observed in the data
(Fig. 6a,d). Light was attenuated
principally by water and, to a lesser
extent by CDOM, which attenuate at
the red and blue end of the light
spectrum, respectively (Kirk 1994,
Pope & Fry 1997, Kitidis et al. 2006).
Attenuation by phytoplankton also
modified the available light spectrum,
but was a rela-
M. Zubkov
tively minor component of the light attenuation in this
oligotrophic ocean.
The normalised light absorption spectra for the entire
phytoplankton community at 2 depths in the model compared
well to observations made during AMT-15 (Fig. 6b,e). Most
importantly, the change in bulk community phytoplankton
absorption spectra with depth was associated with changes in
the wavelengths of available light through the water column.
The shift in relative light absorption, notably from ~510–575 to
450–500 nm, between the surface and deep phytoplankton
populations reflects utilisation of the available light field and
results from contrasts in
Hickman et al.: Modelling phytoplankton communities
9
0–3.5400 500 600 700 400 500 600 700
1.5 x 10–3
(f)
0
0.01
(e)
0 10 E (550):E
(490)
0.01
Surface
DCM
3.5 x 10–3
0 (c)
(b)
(a)
300
400 500 600 700 –1.5
0
400 500 600 700
00 1
(d)
300
Wavelength (nm)
Fig. 6. (a,b,c) Data compared to (d,e,f) model outcomes. (a,d) ratio of irradiance E at 550 to 490 nm through the water column; (b,e) light absorption by the
entire phytoplankton community near the surface (data = 20 m, model = 25 m) and the depth of the deep chl a maximum (DCM; data = 170 m, model =
177.5 m) normalised by the area of the spectra in each case; (c,f) difference in the (normalised) phytoplankton light absorption spectra at the DCM
compared to at the surface. Positive values in (d) and (f) indicate an increase in relative light absorption at the DCM compared to the surface, whereas
negative values indicate a decrease; the area under the curve is equal to 0 such that the y-axis is a relative scale. Data collected during AMT-15 on 17
October 2004, corresponding to Figs. 1c & 5 (optics data provided by G. Moore and L. Hay). Model outcomes are for the ensemble mean (solid
lines) ±SD (dashed lines)
community structure (Fig. 6c,f). This response in the model
was not pre-assigned, but depends on the emergent
phytoplankton community and their combined light absorption
characteristics. Shifts in the wavelengths of phytoplankton light
absorption in response to the available irradiance are indicative
of chromatic adaptation and selection for phytoplankton in the
spectral light gradient (Bidigare et al. 1990a, Lutz et al. 2003,
Hickman et al. 2009).
Variance between ensemble members in the difference in relative
light absorption between the surface and deep populations at 525
to 575 nm in the model (Fig. 6f) reflects whether types with Euk
or Syn spectra dominated towards the surface (Fig. 4), although
on average, the presence of types with Syn spectra towards the
surface led to a decrease in relative light absorption at 550 nm
with depth in the ensemble mean (Fig. 6f). Other contrasts in the
deep and shallow spec-
Fig. 7. Vertical profiles of carbon biomass for the ensemble mean (solid
lines) ±SD (dashed lines) where phytoplankton types are grouped based on
their light absorption properties. (a) Outcome from the full model as in Fig.
5; (b) as in (a), but where all phytoplankton types were assigned the same
spectral light absorption properties (specifically, all types were assigned
light absorption spectra obtained from the mean, at each wavelength, of the
HLPro, LLPro, Euk and Syn spectra shown in Fig. 2. Mean spectra were
obtained for both the light absorption by all pigments, and the light
absorption by photosynthetic pigments). Subsequently, as indicated by
DCM - Surface (relative scale)
Light absorption (relative scale)
Depth (m)
quotation marks, coloured lines in (b) represent the absorption spectra that
would have been assigned for each phytoplankton type in the full model
run, and thus direct comparison of (a) and (b) illustrates the role of the
different light absorption spectra in
forming the vertical distributions
10
Mar Ecol Prog Ser 406: 1–17, 2010
tra between the data and the model, notably at 525 and 675 nm
(Fig. 6b,c,e,f), likely resulted from our simplified community
composition, although it is unlikely that phytoplankton absorption
at the red part of the light spectrum plays a strong role in
chromatic adaptation due to the absence of red light in most of the
water column.
Given the broad agreement between the modelled and observed
phytoplankton distributions and photophysiology, what insights
can the model provide into the key factors driving the community
structure? In particular, is spectral irradiance, and hence
chromatic adaptation, important for the selection of phytoplankton
types in the model?
Drivers of phytoplankton selection
To assess the importance of different traits in forming the
community structures, we considered the consistencies and
differences between outcomes of the ensemble members (Fig. 4),
as well as the ensemble mean (Fig. 7a). A suite of thought
experiments was conducted to illustrate the effect of each of the
temperature, nutrient and spectral light dependencies on the
distribution of phytoplankton types in turn. This suite of
experiments is described fully in Appendix 1. In addition, to focus
on the role of the spectral light absorption properties, we
considered an additional idealised experiment whereby the same
20 ensemble members as in the full model were re-run, this time
with all phytoplankton types assigned identical light absorption
spectra (Fig. 7a,b). The absorption spectrum assigned to all
phytoplankton types in this idealised experiment was the mean of
the representative spectra in Fig. 2 (at each wavelength, and for
spectra for both all-pigments and only photosynthetic pigments).
We start by considering the role of the spectral light
dependence in determining the distribution of phytoplankton
types that utilise nitrate and those that do not, and then explored
the vertical gradients in community structure within these 2 broad
groups.
assigned identical light absorption spectra (Fig. 7b), the nitrateand non-nitrate-using types coexisted throughout most of the
water column at a ratio consistent with that shown for the full
model (Fig. 7a). However, at the DCM in the full model (Fig. 7a),
the biomass for the phytoplankton types that used nitrate (eeuk)
was less than for types that did not (e), whilst in the idealised
experiment, the relative biomasses of these 2 groups were similar
(Fig. 7b). Thus, the spectral absorption properties of LLPro
appear to be advantageous at depth, compared to the absorption
properties of Euk. It follows that the nutrient dependence was the
dominant factor determining the coexistence between the nitrateand non-nitrate-using phytoplankton types, but light absorption
properties also contributed to the dominance of the eLLPLLPat the
DCM. Specifically, the lower possible Ksatvalue assigned to the
phytoplankton types that do not use nitrate (trading off the cost of
reducing nitrate against general nutrient affinity) was the
first-order factor determining the coexistence, whilst LLPro
absorption properties provided a second-order advantage in the
deeper, bluer, part of the water column.
Role of light absorption properties for phytoplankton that do
not use nitrate
The non-nitrate-using phytoplankton types differed according
to their light absorption properties (and photoinhibition) as well as
their ability to utilise NO, or to use NHalong with NH442only (Fig.
3). Sensitivity tests showed that without photoinhibition,
phytoplankton types with LLPro absorption spectra out-competed
those with HLPro spectra at all depths. Thus, spectral light
absorption properties alone were not sufficient to separate eLLPand
eHLPin the light gradient. Including photoinhibition for all
phytoplankton types with LLPro absorption properties,
representing a trade-off for their efficient light harvesting,
obtained realistic vertical distributions of eLLPand eHLP(Fig. 5d).
When implemented in the model, photoinhibition
had a dominant affect on the vertical gradients of the
phytoplankton types that did not use nitrate, causing
dominance of eHLPand exclusion of etowards the surface (Figs.
4 & 5e). At depth, the competitive exclusion of eHLPby eLLPLLP
Light absorption properties and utilisation of nitrate
Nitrate-using (esyn
and eeuk
H
and eLLP
L
near the surface and eLLPHLPat depth occurred in all ensemble members of the full model, such that these features of the
P
community structure were robust even given any combination of initialised Toptand Ksatvalues (Figs. 4 & 5e).
) and non-nitrate-using (e) phytoplankton types coexisted
occurred due to the advantageous light absorption properties of
throughout the water column in all ensemble members of the
LLPro (Figs. 4 & 5e), as confirmed by comparison to the
full model (Figs. 4 & 5d). The relative biomasses of these 2
idealised experiment where there was no competition for
groups were consistent throughout much of the water column,
spectral irradiance (Fig. 7a,b). The dominance of e
although the ratio of non-nitrate- to nitrate-using types
increased at the DCM. In the idealised experiment, where
phytoplankton types were
Hickman et al.: Modelling phytoplankton communities
and Ksat
opt
Laboratory and field observations suggest that photoinhibition is
important for the general absence of some low-light
Prochlorococcus ecotypes in surface waters (Moore & Chisholm
1999, Zinser et al. 2007), although there are likely to be other or
additional trade-offs important to the niche separation of these
organisms (Zinser et al. 2007). Culture studies have shown that a
low-light Prochlorococcus ecotype (SS120) is more susceptible
to photoinhibition than a high-light strain due to a relatively slow
repair rate of photodamaged photosystem-II’s (Six et al. 2007).
All Prochlorococcus ecotypes may be more susceptible to
photoinhibition than other phytoplankton due to the presence of
divinyl chlorophylls, which are considered to be more expensive
to repair than the monovinyl chlorophyll forms (Tomo et al.
2009). The low-light Prochlorococcus, with a high divinyl chl b:
divinyl chl a ratio, are adapted to highly efficient utilisation of
blue light at depth, while the relatively less efficient light
absorption properties of high-light Prochlorococcus, with a low
divinyl chl b:divinyl chl a ratio, presumably reflects a balance
between light absorption for photosynthesis and minimising the
risk of photodamage.
opt and K
syn
syn
euk
sat
euk
sat
and eeuk
tween esyn
In the model, the utilisation of different nitrogen sources was also
important for the vertical distributions of phytoplankton types that
did not use nitrate, representing Prochlorococcus. At the DCM,
and thus at the nutricline, eLLPthat used NO2as well as NHhad a
competitive advantage compared to those that only used NH44. In
contrast, near the surface where NO2
concentrations were low, eHLP
2
11
was due to the stochastic assignment of Topt
for the phytoplankton types within each group, and this difference
was not significant. It was therefore the light absorption properties
that had resulted in the vertical gradients of these phytoplankton
types in the full model (Figs. 5d & 7a). The phytoplankton types
with Syn absorption spectra (e) were not successful at the depth of
the DCM in any ensemble member due to their competitive
exclusion by those with Euk absorption spectra (e; Fig. 4),
something that could not be overcome by any co-assigned
combination of Tand Kvalues within the given ranges (Figs. 4 &
5e). Towards the surface, however, the competition bewas more complicated. Whether a phytoplankton type with Syn
or Euk absorption properties dominated near the surface varied
between ensemble members (Fig. 4), and thus depended on the
values of Tfor the corresponding phytoplankton types. Overall,
however, types with Syn absorption properties had a
competitive advantage over those with Euk spectra as
evidenced by their higher biomass in the ensemble mean (Fig.
5). At 25 m in the ensemble mean, the biomass of ewas
significantly greater than the biomass of eat the 95%
confidence level (according to a t-test).
that utilised NO2
4
2
4
The light absorption properties affect the growth rate according to Eq. (7). The value of the
summed component, and therefore ΛI j, increases when the light absorption spectrum is similar to
the wavelength composition of available light (Sathyendranath & Platt 2007). For the light field
of the ensemble mean state, the value of ΛI j– 1 was 20% higher for phytoplankton types with Syn
compared to Euk absorption spectra at 25 m, but the growth rate (s) was only 1% greater (given
the nutrient fields from the ensemble mean and assuming that all other characteristics were equal:
Topt= 22°C and Ksat= 0.01). This small difference in growth rate is partly due to compensation by
photoacclimation, which results in a different chl a to carbon ratio between the phytoplankton
types (0.025 and 0.030 g:g for the type with Syn and Euk absorption spectra, respectively). In
contrast, the value of ΛI jand the growth rate were both 5% higher for the phyto-
as well as
NHas well as NHdid not have a significant advantage. This
preference for reduced nitrogen source is consistent with culture
studies showing that low-light Prochlorococcus ecotypes
assimilate NO, whilst cultured high-light ecotypes have lost the
ability to use NO(Moore et al. 2002, Rocap et al. 2003). The
model therefore supports the view that this trait is also important
for niche partitioning of different Prochlorococcus ecotypes on
vertical nutrient gradients (Moore et al. 2002, Bragg et al. 2010).
Role of light absorption properties for phytoplankton that use
nitrate
The group of phytoplankton types that use nitrate consisted of
phytoplankton with Syn and Euk spectral absorption properties.
In the idealised experiment without competition for spectral
irradiance (Fig. 7b), the biomass of the phytoplankton types
that had, in the full model, been assigned Syn and Euk
absorption properties (the red and green lines in Fig. 7b)
followed the same vertical gradient; their slight difference in
absolute biomass
In an additional idealised experiment where all of their other
traits were equal, phytoplankton types with absorption spectra
of Syn and Euk coexisted towards the surface (see Appendix 1:
Test 1, Fig. A1b for the full description of this experiment).
This coexistence occurs due to their utilisation of different
components of the light spectrum (Stomp et al. 2004, 2007),
resulting in similar light-dependent growth rates. In the full
model, the spectral absorption properties of these
phytoplankton types thus play a less dominant role, allowing
other traits to ‘tip-the-balance’ and be more significant in
forming the resulting distributions.
12
Mar Ecol Prog Ser 406: 1–17, 2010
sat ties
could survive at depth, were it not for their competitive
exclusion by eeuk. The slight advantage of phytoplankton typ
with
Syn compared to Euk absorption properties towards the surf
implies some light limitation of growth. The precise ratio of
biomass of these types (such as in Appendix 1: Test 1, Fig.
is likely to result from a subtle interplay between light absor
and mixing (Stomp et al. 2007).
plankton types with Euk compared to Syn absorption spectra at
170 m, where the ratio of chl a to carbon was near maximal.
Thus, the light absorption properties had a greater influence on
growth rate at depth, and led to competitive exclusion of esynby
ein the full model, even for any combination of Topteukand Kvalues
within the prescribed ranges (Figs. 3 & 4). The importance of the
light absorption properties at depth reflects the fact that ΛI
jsatdetermines the light-limited slope of the photosynthesis versus
irradiance curve and hence light-limited growth rate (Eqs. 4 and
6). The absence of phytoplankton types with Syn absorption
spectra at depth in the model is consistent with Synechococcus
being light-limited due to their spectral light requirements (Wood
1985). Our model indicated that phytoplankton types with Syn
absorption proper-
sat
For the example ensemble member in Fig. 4a, the assigne
values of Ksatand Twere compared for the successful versus
unsuccessful phytoplankton types (Fig. 8a,b). With 1000
phytoplankton types initialised, the assigned values of Toptop
K– 5 clearly covered parameter space, although only a minor
persisted with a biomass greater than 10µM P. For the nutri
sensitivity, phytoplankton types with
out-competed those with higher K
values (Fig. 8a,b), but this selection did not lead to vertical gradients in community
structure (see Appendix 1: Test 2 for further evidence of this point). This successful
competition can be explained by the openended Michaelis-Menten function for the
nutrientgrowth response, whereby, for any given nutrient concentration, a
phytoplankton type with lower Kvalue yields a higher γN jand therefore PC m,jsat, than
one with a higher Ksatvalue. In this configuration, when Kwas assigned randomly for
the given ranges, there was no trade-off for a lower Ksatsat. This is a simplification of
the complex traits and trade-offs that govern nutrient uptake, for example, causing the
maximum uptake rate to co-vary with K(Litchman et al. 2007). The selection for
Ksatsatmay be less important in nutrient replete regimes (Dutkiewicz et al. 2009).
Role of Topt and K
low values of K
sat
sat
(HLPro) absorption spectra (black) and low-light Prochlorococcus
(LLPro) absorption spectra (blue), and (b) phytoplankton types with
Synechococcus (Syn) absorption spectra (green) and eukaryote (Euk)
absorption spectra (red). Values of Topt– 5 against the temperature at the
depth of the biomass maxima (at Month 10) for successful phytoplankton
types with (c) HLPro absorption spectra (black) and LLPro absorption
spectra (blue), and (d) Syn absorption spectra (green) and Euk absorption
spectra (red). In (c) and (d), only the successful phytoplankton types are
included; the 1:1 line is also shown. Successful phytoplankton types are
those with biomass above 10µM P at Month 10, and subsequently those
types visible in Fig. 4a
Fig. 8. Initialised parameter values for the example ensemble member
shown in Fig. 4a. Values of the optimum temperature for growth, T(°C),
against the nutrient half-saturation constant, Ksatopt(µM P), for all
initialised phytoplankton types (dots) and successful phytoplankton types
(diamonds), for (a) phytoplankton types with high-light Prochlorococcus
For the temperature sensitivity, phytoplankton types had a
competitive advantage at the depth where the in situ temperature
was similar to their assigned value of Topt(Fig. 8c,d). Since the
temperature-growth function has upper and lower limits (Eq. 10),
it follows that a phytoplankton type has a competitive advantage
within a narrow range of in situ temperatures. In contrast to Ksat,
temperature dependence was therefore important for vertical
gradients in community structure, and selection based on
temperature did not reduce the number of viable phytoplankton
types (see Appendix 1: Test 3 for further evidence). Again, there
was no associated trade-off for the temperature dependence in the
model.
The stochastic framework employed in the model (Fig. 3)
results in both Ksatand Toptvalues for each successful
phytoplankton type effectively being ‘opti-
Hickman et al.: Modelling phytoplankton communities
mised’ for the environmental conditions. The optimisation of
traits is in accord with the concept of evolutionary selection that
leads to phytoplankton having multiple characteristics that suit
them to a given environment. For example, high-light and
low-light Prochlorococcus ecotypes have corresponding high
and low Topt, respectively (Zinser et al. 2007). The cooptimal
adaptation of multiple traits between Prochlorococcus ecotypes
also extends to their differential nitrogen sources (Moore et al.
2002) and underpins their biogeography (Johnson et al. 2006,
Zinser et al. 2007). However, the combination of temperature and
light dependence clearly involves a complicated interplay
between traits and trade-offs (Zinser et al. 2007).
The simplified mixing regime in the model may have led to an
overemphasised temperature gradient (Fig. 5a,b), and
subsequently temperature-driven gradients in community
structure above the DCM (Appendix 1: Fig. A1d). However, the
selection for temperature occurs within, rather than between,
spectral classes such that it does not influence our assessment of
the importance of chromatic adaptation.
DISCUSSION
A stochastic, ‘self-organising’ model of marine phytoplankton
was employed to investigate the role of chromatic adaptation in
determining the vertical organisation of the phytoplankton
communities in the oligotrophic oceans. We have implemented
parameterisations of photoacclimation along with
wavelengthdependent light absorption and radiative transfer in the
model. One thousand phytoplankton types were each initialised
with a unique set of physiological characteristics, including
spectral light absorption properties, and the phytoplankton
community was allowed to self-assemble through in silico
‘survival of the fittest’.
The model was applied for a vertical profile in the South Atlantic
Gyre, representative of the oligotrophic environment covering
much of the ocean’s midlatitudes. The emergent profiles of
phytoplankton biomass, chl a and vertical gradients in community
structure of the phytoplankton were qualitatively consistent with
observations from the Atlantic Meridional Transect, albeit with
some contrasts in absolute values. Following a number of model
thought experiments, the skill of the model in reproducing the
community structure was strongly dependent on chromatic
adaptation; the model’s self-assembled phytoplankton community
reflects a selective pressure from the photophysiology and
pigments. Our demonstration of the importance of chromatic
adaptation in an oceanic simulation is consistent with
observational (e.g. Bidigare et al. 1990a, Ting et al. 2002, Lutz et
al. 2003) and idealised model
13
studies (Stomp et al. 2004, Sathyendranath & Platt 2007). In the
model, organisation of the phytoplankton
community was strongly affected by the inorganic nitrogen
source, which segregated the water column into 2 broad groups:
phytoplankton types that used nitrate (Synechococcus and
eukaryotes) and those that did not (Prochlorococcus, or a subset
thereof). The competition for spectral irradiance had a secondary
influence on the coexistence between these 2 groups, most
notably that phytoplankton types with absorption properties of the
low-light Prochlorococcus had an advantage over those with
absorption properties of the picoeukaryote at the DCM. Whether
Prochlorococcus or eukaryotes dominate at the base of the
euphotic zone thus depends on the co-availability of light and
nutrients. Within the 2 broad groups controlled by the inorganic
nitrogen source, pigments and photophysiology played a
dominant role in structuring the community. Their spectral light
requirements restricted Synechococcus to a surface water
habitat due to competitive exclusion by eukaryotes at depth. In the
surface, however, these 2 phytoplankters could coexist depending
on their other traits. The spectral light requirements of the highand low-light Prochlorococcus types did not lead to a vertical
separation. Instead, realistic habitat organisation was only
achieved when photoinhibition was imposed as a trade-off to the
highly efficient light absorption of the low-light types. This role
of photoinhibition for the absence of low-light Prochlorococcus
ecotypes near the surface is consistent with culture and field
studies (Moore & Chisholm 1999, Six et al. 2007, Zinser et al.
2007).
Multiple traits, here specifically the growth rate dependencies
on light, nutrients and temperature, were co-optimal for the depths
where phytoplankton types occurred in the water column. Thus,
while optical properties play a role in the structuring of habitat,
they represent just one of several factors that have coevolved to
be optimal in a particular environment and together shape the
niches of different phytoplankton.
The model study made a number of simplifying assumptions in
order to focus on the role of chromatic adaptation.
Prochlorococcus ecotypes that could utilise nitrate (Casey et
al. 2007), other size classes, or additional traits such as grazing
defence, mixotrophy or mobility that potentially influence
niche partitioning in the real ocean, were not explicitly
included in the model. There were also simplifying
assumptions about the chemical environment (such as a fixed
uptake ratio for nutrients and no iron dependence) and the
parameterisations for growth dependencies (such as the Eppley
curve for temperature and Michaelis-Menten form for
nutrient-dependent growth; Follows et al. 2007, Dutkiewicz et
al. 2009). Although these simplifications may have influenced
the resulting absolute
14
Mar Ecol Prog Ser 406: 1–17, 2010
biomass values, they would not qualitatively alter the effect of
spectral irradiance on species selection.
For this initial study, 4 phytoplankton types were chosen that
had distinctly different light absorption spectra. The effects of
photoacclimation and pigment packaging on the absorption
spectra were not described, but were implicit in the culture
measurements (Fig. 2). We subsequently assume that the 4 chosen
types have absorption properties similar to the species within the
groups they represent (i.e. Synechococcus, high-light and
low-light Prochlorococcus and eukaryotes), and that spectral
effects of acclimation are negligible compared to the difference in
absorption properties between the groups. This assumption is
reasonable, noting particularly that the picoeukaryote population
in the oligotrophic regions also consists of flagellates that are of
similar size and contain pigments that absorb light at similar
wavelengths to our representative eukaryote, and that
Synechococcus ecotypes with larger amounts of other
phycobilins than are contained in our phycoerythrobilin-rich
representative also occur in the oligotrophic ocean (Scanlan et al.
2009). While our simplified choices for the representative
absorption spectra could have contributed to the mismatch in
absolute biomass of the different types (Fig. 5c,d), as well as the
community light absorption properties (Fig. 6b,c,e,f), between the
model and the data, the model successfully reproduced the
observed vertical gradients in community structure and revealed a
clear role of chromatic adaptation that is consistent with
observations. The importance of other phycobilins (phycourobilin
and phycocyanobilin) within Synechococcus, or the range of
pigment types within different eukaryotes could be explored in
future studies.
Our results reveal co-optimality of multiple traits for a given
niche, consistent with the evolutionary view. A key next step is
exploring the interdependencies of traits and their physiological
trade-offs (e.g. Litchman et al. 2007). For example, there is an
interesting trade-off between light and nitrogen requirement in
Synechococcus. Phycobilisomes are particularly rich in nitrogen
(Raven 1984), creating a high nitrogen demand for the cells and,
presumably, a preference for a nitrogen (and therefore
nitrate)-rich environment. At the same time, the spectral
dependence of absorption by these phycobilisomes favours a
surface water habitat. The tension between these selective
pressures may be important for the distribution of
Synechococcus in the very oligotrophic central gyres, where
they occur in near surface waters but do not dominate the
phytoplankton biomass (Zubkov et al. 1998, 2000, Heywood et al.
2006). A complex ecosystem model, such as that used here, could
provide a useful framework for future targeted studies into the
interdependences of traits and trade-offs, and their role for
community ecology.
In summary, we have demonstrated the importance of chromatic
adaptation for shaping phytoplankton communities in the South
Atlantic Gyre. The phytoplankton growth dependencies on light,
nutrients and temperature were co-optimal, supporting the view
that multiple traits have co-evolved to define an organism’s niche.
Although we have focussed on an oligotrophic environment, the
processes and interactions explored here underpin phytoplankton
selection and thus have a wider relevance to the global ocean.
Acknowledgements. We thank D. Suggett and L. Moore for providing
culture data; J. Heywood and M. Zubkov for analytical flow cytometry
measurements; L. Hay and G. Moore for spectral optical profiler data; the
wider Atlantic Meridional Transect (AMT) community for providing
ancillary data; and the officers and crew of RRS ‘Discovery’ during
AMT-15. We also thank W. Gregg for valuable discussion on the optical
framework and 5 anonymous referees for their insightful comments. This
work was jointly funded through the UK Natural Environment Research
Council (NERC) (grant NE/ F002408/1), NASA (grant NNX09AE44G)
and the Gordon and Betty Moore Foundation. The study was also
supported by NERC through the AMT consortium (NER/O/S/2001/
00680) and is contribution number 186 of the AMT programme.
Participation of A.E.H. in AMT-15 was funded by a NERC PhD
studentship.
LITERATURE CITED
Babin M, Morel A, Claustre H, Bricaud A, Kolber Z, Falkowski PG
(1996) Nitrogen- and irradiance-dependent variations of the maximum
quantum yield of carbon fixation in eutrophic, mesotrophic and
oligotrophic marine systems. Deep-Sea Res I 43:1241–1272
Barlow RG, Aiken J, Holligan PM, Cummings DG, Maritorena S, Hooker
S (2002) Phytoplankton pigment and absorption characteristics along
meridional transects in the Atlantic Ocean. Deep-Sea Res II 47:637–660
Bidigare RR, Marra J, Dickey TD, Iturriaga R, Baker KS, Smith RC, Pak
H (1990a) Evidence for phytoplankton succession and chromatic
adaptation in the Sargasso Sea during spring 1985. Mar Ecol Prog Ser
60:113–122
Bidigare RR, Ondrusek ME, Morrow JH, Kiefer DA (1990b) In-vivo
absorption properties of algal pigments. Proc SPIE 1302:290
Bragg JG, Dutkiewicz S, Jahn O, Follows MJ, Chisholm SW (2010)
Modeling selective pressures on picocyanobacterial nitrogen use in the
global ocean. PLoS ONE 5:e9569, doi:10.1371/journL.PONE.0009569
Bricaud A, Stramski D (1990) Spectral absorption coefficients of living
phytoplankton and nonalgal biogenous matter: a comparison between the
Peru upwelling area and the Sargasso Sea. Limnol Oceanogr 35:562–582
Casey J, Lomas MW, Mandecki J, Walker DE (2007) Prochlorococcus
contributes to new production in the Sargasso Sea deep chlorophyll
maximum. Geophys Res Lett 34:L10604, doi:10.1029/2006GL028725
Chisholm SW (1992) Phytoplankton size. In: Falkowski PG, Woodhead
AD (eds) Primary productivity and biogeochemical cycles in the sea.
Plenum Press, New York, NY, p 213–237
Conkright ME, Garcia HE, O’Brian TD, Locarnini RA, Boyer
Hickman et al.: Modelling phytoplankton communities
TP, Stephens C, Antonov JI (2002) World ocean atlas 2001, Vol 4.
Nutrients. In: Levitus S (ed) NOAA Atlas NESDIS 52. US
Government Printing Office, Washington, DC
Dutkiewicz S, Follows MJ, Parekh P (2005) Interactions of the iron and
phosphorus cycles: a three-dimensional model study. Glob Biogeochem
Cycles 19:GB1021, doi:10.1029/ 2004GB002342
Dutkiewicz S, Follows MJ, Bragg J (2009) Modeling the coupling of
ocean ecology and biogeochemistry. Glob Biogeochem Cycles
23:GB4017, doi:10.1029/2008GB003405
Eppley RW (1972) Temperature and phytoplankton growth in the sea.
Fish Bull 70:1063–1085
Follows MJ, Dutkiewicz S, Grant S, Chisholm SW (2007) Emergent
biogeography of microbial communities in a model ocean. Science
315:1843–1846
Geider RJ, MacIntyre HL, Kana TM (1997) Dynamic model of
phytoplankton growth and acclimation: responses of the balanced growth
rate and the chlorophyll a:carbon ratio to light, nutrient-limitation and
temperature. Mar Ecol Prog Ser 148:187–200
Gregg WW, Casey NW (2007) Modeling coccolithophores in the global
oceans. Deep-Sea Res II 54:447–477
Heywood JL, Zubkov MV, Tarran GA, Fuchs BM, Holligan PM (2006)
Prokaryoplankton standing stocks in oligotrophic gyre and equatorial
provinces of the Atlantic Ocean: evaluation of inter-annual variability.
Deep-Sea Res II 53:1530–1547
Hickman AE, Holligan PM, Moore CM, Sharples J, Krivtsov V, Palmer
MR (2009) Distribution and chromatic adaptation of phytoplankton within
a shelf sea thermocline. Limnol Oceanogr 54:525–536
Jeffrey SW, Mantoura REC, Wright SW (1997) Phytoplankton pigments
in oceanography: guidelines to modern methods. UNESCO, Paris
Johnson ZI, Zinser ER, Coe A, McNulty NP, Woodward EMS, Chisholm
SW (2006) Niche partitioning among Prochlorococcus phenotypes along
ocean-scale environmental gradients. Science 311:1737–1740
Kettle H, Merchant CJ (2008) Modeling ocean primary production:
sensitivity to spectral resolution of attenuation and absorption of light.
Prog Oceanogr 78:135–146
Kirk JTO (1994) Light and photosynthesis in aquatic ecosystems.
Cambridge University Press, Cambridge
Kitidis V, Stubbins AP, Uher G, Upstill Goddard RC, Law CS, Woodward
EMS (2006) Variability of chromophoric organic matter in surface waters
of the Atlantic Ocean. Deep-Sea Res II 53:1666–1684
Kyewalyanga MN, Platt T, Sathyendranath S, Lutz VA, Stuart V (1998)
Seasonal variations in physiological parameters of phytoplankton across
the North Atlantic. J Plankton Res 20:17–42
Ledwell JR, Watson AJ, Law CS (1993) Evidence for slow mixing across
the pycnocline from an open-ocean tracerrelease experiment. Nature
364:701–703
Litchman E, Klausmeier CA, Schofield OM, Falkowski PG (2007) The
role of functional traits and trade-offs in structuring phytoplankton
communities: scaling from cellular to ecosystem level. Ecol Lett
10:1170–1181
Lutz VA, Sathyendranath S, Head EJH, Li WKW (2003) Variability in
pigment composition and optical characteristics of phytoplankton in the
Labrador Sea and the Central North Atlantic. Mar Ecol Prog Ser 260:1–18
Moore LR, Chisholm SW (1999) Photophysiology of the marine
cyanobacterium Prochlorococcus: ecotypic differences among cultured
isolates. Limnol Oceanogr 43: 628–638
Moore LR, Goericke R, Chisholm S (1995) Comparative phys-
15
iology of Synechococcus and Prochlorococcus: influence of light
and temperature on growth, pigments, fluorescence and absorptive
properties. Mar Ecol Prog Ser 116: 259–275
Moore LR, Rocap G, Chisholm SW (1998) Physiology and molecular
phylogeny of coexisting Prochlorococcus phenotypes. Nature
393:464–466
Moore LR, Post AF, Rocap G, Chisholm SW (2002) Utilization of
different nitrogen sources by the marine cyanobacteria Prochlorococcus
and Synechococcus. Limnol Oceanogr 47:989–996
Partensky F, Blanchot J, Lantoine F, Neveux J, Marie D (1996) Vertical
structure of picophytoplankton at different trophic sites of the tropical
northeastern Atlantic Ocean. Deep-Sea Res I 43:1191–1213
Pope RM, Fry ES (1997) Absorption spectrum (380–700 nm) of pure
water. II. Integrating cavity measurements. Appl Opt 36:8710–8723
Rabouille S, Edwards CA, Zehr JP (2007) Modelling the vertical
distribution of Prochlorococcus and Synechococcus in the North
Pacific Subtropical Ocean. Environ Microbiol 9:2588–2602
Raven JA (1984) A cost-benefit analysis of photon absorption by
photosynthetic unicells. New Phytol 98:593–625
Rocap G, Larimer FW, Lamerdin JW, Malfatti S and others (2003)
Genome divergence in two Prochlorococcus phenotypes reflects oceanic
niche differentiation. Nature 424: 1042–1047
Sathyendranath S, Platt T (2007) Spectral effects in bio-optical control on
the ocean system. Oceanologia 49:5–39
Scanlan DJ, Ostrowski M, Mazard S, Dufresne A and others (2009)
Ecological genomics of marine picocyanobacteria. Microbiol Mol Biol
Rev 73:249–299
Six C, Finkel ZV, Irwin AJ, Campbell DA (2007) Light variability
illuminates niche-partitioning among marine picocyanobacteria. PLoS
ONE 2:e1341, doi:10.1371/journal. pone.0001341
Stephens C, Antonov JI, Boyer TP, Conkright ME, Locarnini RA,
O’Brian TDO, Garcia HE (2002) World ocean atlas 2001, Vol 1.
Temperature. In: Levitus S (ed) NOAA Atlas NESDIS 49. US
Government Printing Office, Washington, DC
Stomp M, Huisman J, de Jongh F, Veraart AJ and others (2004) Adaptive
divergence in pigment composition promotes phytoplankton biodiversity.
Nature 432:104–107
Stomp M, Huisman J, Vцrцs L, Pick FR, Laamanen M, Haverkamp T,
Stal LJ (2007) Colourful coexistence of red and green picocyanobacteria
in lakes and seas. Ecol Lett 10:290–298
Suggett DJ, MacIntyre HL, Geider RJ (2004) Evaluation of biophysical
and optical determinations of light absorption by photosystem II in
phytoplankton. Limnol Oceanogr Methods 2:316–332
Suggett DJ, Moore CM, Hickman AE, Geider RJ (2009) Interpretation of
fast repetition rate (FRR) fluorescence: signatures of phytoplankton
community structure versus physiological state. Mar Ecol Prog Ser
376:1–19
Ting CS, Rocap G, King J, Chisholm SW (2002) Cyanobacterial
photosynthesis in the oceans: the origins and significance of divergent
light harvesting strategies. Trends Microbiol 10:134–142
Tomo T, Akimoto S, Ito H, Tsuchiya T, Fukuya M, Tanaka A, Mimuro
M (2009) Replacement of chlorophyll with divinyl chlorophyll in the
antenna and reaction centre complexes of the cyanobacterium
Synechocystis sp. PCC 6803: characterisation of spectral and
photochemical properties. Biochim Biophys Acta 1787:191–200
Mar Ecol Prog Ser 406: 1–17, 201016
Williams RG, Follows MJ (1998) The Ekman transfer of nutrients and
maintenance of new production over the North Atlantic. Deep-Sea Res I
45:461–489
Wood AM (1985) Adaptation of photosynthetic apparatus of marine
ultraphytoplankton to natural light fields. Nature 316:253–255
Ocean. Limnol Oceanogr 52:2205–2220 Zubkov MV, Sleigh MA, Tarran
GA, Burkill PH, Leakey
RJG (1998) Picoplanktonic community structure on an Atlantic
transect from 50°N to 50°S. Deep-Sea Res I 45: 1339–1355
Zubkov MV, Sleigh MA, Burkill PH, Leakey RJG (2000) Picoplankton
community structure on the Atlantic Meridional Transect: a
comparison between seasons. Prog Oceanogr 45:369–386
Zinser ER, Johnson ZI, Coe A, Karaca E, Veneziano D, Chisholm SW
(2007) Influence of light and temperature on Prochlorococcus phenotype
distributions in the Atlantic
Appendix 1. Thought experiments
and Topt
sat
opt
sat
opt
sat
opt
euk
opt
T
o
f
u
r
t
h
e
r
e
x
a
m
i
n
e
t
h
e
i
m
p
o
r
t
a
n
c
e
o
f
l
i
g
h
t
a
b
s
o
r
p
t
i
o
n
s
p
e
c
T
e
s
t
1
.
S
a
m
e
K
s
a
t
f
o
r
a
l
l
p
h
y
t
o
p
l
a
n
k
Fig. A1. (a) Vertical profiles of carbon biomass for all phytoplankton
ty
t thought
according to their light absorption properties. (b to e) Results of
re-run with chosen characteristics assigned to be the same for oall phytop
n (that,
enlarged region 0 to 175 m for a subset of 20 phytoplankton types
vertical gradients caused by different optimum temperature (T) values.
all phytoplankton types, the light absorption spectrum was thet mean (at
y (Syn)
low-light Prochlorococcus (LLPro), Euk and Synechococcus
p
absorption by all pigments, and the light absorption by photosynthetic
p
e been
lines in (c) to (e) represent the absorption spectra that would have
s
thus allow direct comparison with (a). Ksat: half-saturation constant;
Dif
,
model run); Abs.: absorption spectra
r
e
a
l
i
s
t
i
c
a
b
s
o
r
p
t
i
o
n
s
p
e
c
t
r
a
a
s
s
i
g
n
e
d
(
F
i
g
.
Since there was no stochastic assignment for T
In the first experiment, T
and K
and K
Hickm
an et
al.:
Model
ling
phyto
plankt
on
comm
unities
17
Ap
pen
dix
1
(co
nti
nue
d)
ferences in light absorption properties. This result shows that the light
absorption properties allow coexistence of esyn
and eeuktowards the surface, resulting from their utilisation of different
components of the light spectrum (Stomp et al. 2004, 2007). Further, it
reveals an advantage for Syn compared to Euk spectra towards the
surface, given the higher biomass for esyn(Fig. A1b). The dominant
effects of spectral absorption properties and photoinhibition discussed
in the main text also describe the vertical separation of phytoplankton
types with HLPro and LLPro absorption properties in this experiment.
This experiment revealed similar vertical gradients in community
composition as observed for the ensemble mean of the full model (Fig.
5d).
Test 2. Same Topt and absorption spectra for all types, Ksat
In a
final
thoug
ht
experi
ment,
all
phyto
plankt
on
types
had
the
same
(mean
) light
absorp
tion
spectr
a,
whilst
Topt
and
Ksatwe
re
assign
ed
stocha
sticall
y.
This
experi
and corresponding competition on the temperate gradient (Fig. 8c,d),
caused fine-scale vertical gradients in the community structure through
the water column (inset in Fig. A1d). This selection on the temperature
gradient was responsible for the success of numerous different
phytoplankton types within the light resource groupings, for example,
the multiple phytoplankton types with LLPro absorption spectra (i.e.
blue lines) in the full ensemble member (Fig. A1a). Selection due to
the temperature dependence did not reduce the number of successful
phytoplankton types.
The vertical gradients in community structure observed in the full
model outcome were thus dependent on the temperature
dependence; phytoplankton types that were successful towards the
surface had high values of T, whilst those successful at depth had
lower values of Topt.opt
ment
thus
repres
ents
an
ensem
ble
memb
er that
contri
buted
to the
mean
shown
in Fig.
7b.
Altho
ugh
not
identic
al,
there
were
simila
rities
betwe
en this
experi
ment
(Fig.
A1e)
and
the
ensem
ble
outco
me
from
the
full
model
(Fig.
A1a).
For
examp
le, in
the
ensem
ble
from
the
full
model
(Fig.
A1a),
a
phyto
plankt
on
type
with
Syn
absorp
tion
spectr
a
domin
ated at
the
surfac
e,
whilst
in the
experi
ment
(Fig.
A1e),
this
also
occurr
ed,
albeit
by
chanc
e due
to the
assign
ed
Toptan
d Ksat
val
ues
,
rat
her
tha
n
spe
ctra
l
lig
ht
abs
orp
tio
n
pro
per
ties
.
Si
mil
arl
y,
Ksat
and
Topt
wer
e
opt
ima
l
for
gro
wth
at
the
DC
M
for
the
phy
top
lan
kto
n
typ
e
als
o
assi
gne
d
Eu
k
abs
orp
tio
n
spe
ctra
in
the
exa
mp
le
ens
em
ble
(Fi
g.
A1
a,e)
. It
foll
ow
s
that
in
the
full
exa
mp
le
ens
em
ble
me
mb
er
(Fi
g.
A1
a),
the
suc
ces
sful
phy
top
lan
kto
n
typ
es
wit
h
Eu
k
and
Sy
n
lig
ht
abs
orp
tio
n
pro
per
ties
had
dep
end
enc
ies
on
tem
per
atu
re,
nut
rie
nts
and
lig
ht
that
wer
e
coopt
ima
l.
Suc
ces
sful
phy
top
lan
kto
n
typ
es
in
the
full
mo
del
the
ref
ore
als
o,
on
ave
rag
e,
had
a
coo
pti
mal
co
mb
inat
ion
of
trai
ts.
Test 4. Same absorption spectra for all types, Topt
assigned stochastically (Fig. A1e)
assigned stochastically (Fig. A1c)
In the second thought experiment, all phytoplankton types were
provided with the same (mean) light absorption spectra and a constant
value for Topt(Topt= 22°C), while Ksatwas assigned stochastically. The
line colours in Fig. A1c represent the absorption spectrum and were
assigned for directly corresponding phytoplankton types in the full
model, even though the light absorption properties are now the same
for all types. This enables a comparison to the distributions inthe full
model outcome (Fig. A1c). Accepting that photoinhibition and the
nitrogen resource requirements caused some vertical structure within
the non-nitrate-using phytoplankton types, there were no apparent
vertical shifts in community structure caused by the contrasting values
of K. The selection for low Ksatsatvalues (Fig. 8a,b) strongly influenced
which phytoplankton types were successful and significantly reduced
the number of viable types. However, the successful types were
consistently dominant at all depths. Thus, in the full model, the
selection for Kvalues reduced the number of potentially successful
phytoplankton types (and all successful types had low Ksatsatvalues) but
did not significantly influence the vertical gradients in community
structure.
Test 3. Same Ksat and absorption spectra for all types, Topt
assigned stochastically (Fig. A1d)
In the third thought experiment, all phytoplankton types had the
same (mean) light absorption spectra and the same Ksatvalue (Ksat=
0.015 µM P for all phytoplankton types that used nitrate and 0.01 µM
P for those that did not), while Topt
was assigned stochastically. Variations in the value of Topt,
responsibility: Graham Savidge, Portaferry, UK
Editorial
and Ksat
opt
act on phytoplankton selection. Crucially, the model preferentially In summary, this series of thought exper
selected for phytoplankton with co-optimal traits. Thus, the successful
spectral light properties, Ksatand T
phytoplankton types in the ensemble mean (Fig. 5) had, in addition to
their dependence on spectral irradiance, low half saturation constants
for nutrient uptake and temperature optima suitable for the depth at
which they occurred in the water column.
Submitted: October 3, 2009; Accepted: March 16, 2010
Proofs received from author(s): April 11, 2010